SummerTrains
SummerTrains
请问生成PDF文件后，术语与符号约定那个目录不见了是什么情况？
@JThh thx！ I have setted text_encoder.requires_grad_(True) and text_encoder.train(), but parameter of text-encoder was not changed. I print `text_encoder.text_model.encoder.layers[3].self_attn.q_proj.weight`, it has not been changed!
@JThh And I think, it is that the text-encoder has not been warpped with GeminiAdamOptimizer or other optimizer
@JThh The most important code is `params_to_optimize = ( itertools.chain(unet.parameters(), text_encoder.parameters()) if args.train_text_encoder else unet.parameters() )`, and sand it into torch.optimizer. But it's different with GeminiAdamOptimizer. The key point is...
> Would you please change [this](url) `__init__` method to possibly be: > > ``` > def __init__(self, models: Union[List[torch.nn.Module], torch.nn.Module] , **defaults: Any) -> None: > if isinstance(models, list): >...
@JThh hello, how is going? When will you finish this problem?